1 /* Asynchronous subprocess implementation for UNIX
2 Copyright (C) 1985, 1986, 1987, 1988, 1992, 1993, 1994, 1995
3 Free Software Foundation, Inc.
4 Copyright (C) 1995 Sun Microsystems, Inc.
5 Copyright (C) 1995, 1996 Ben Wing.
7 This file is part of XEmacs.
9 XEmacs is free software; you can redistribute it and/or modify it
10 under the terms of the GNU General Public License as published by the
11 Free Software Foundation; either version 2, or (at your option) any
14 XEmacs is distributed in the hope that it will be useful, but WITHOUT
15 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
16 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
19 You should have received a copy of the GNU General Public License
20 along with XEmacs; see the file COPYING. If not, write to
21 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
22 Boston, MA 02111-1307, USA. */
24 /* This file has been Mule-ized except for `start-process-internal',
25 `open-network-stream-internal' and `open-multicast-group-internal'. */
27 /* This file has been split into process.c and process-unix.c by
28 Kirill M. Katsnelson <kkm@kis.ru>, so please bash him and not
29 the original author(s) */
31 /* The IPv6 support is derived from the code for GNU Emacs-20.3
32 written by Wolfgang S. Rupprecht */
36 #if !defined (NO_SUBPROCESSES)
38 /* The entire file is within this conditional */
53 #include "file-coding.h"
60 #include "syssignal.h" /* Always include before systty.h */
66 * Implementation-specific data. Pointed to by Lisp_Process->process_data
69 struct unix_process_data
71 /* Non-0 if this is really a ToolTalk channel. */
72 int connected_via_filedesc_p;
73 /* Descriptor by which we read from this process. -1 for dead process */
75 /* Descriptor for the tty which this process is using.
76 -1 if we didn't record it (on some systems, there's no need). */
78 /* Name of subprocess terminal. */
80 /* Non-false if communicating through a pty. */
84 #define UNIX_DATA(p) ((struct unix_process_data*)((p)->process_data))
87 /* The file name of the pty opened by allocate_pty. */
89 static char pty_name[24];
94 /**********************************************************************/
95 /* Static helper routines */
96 /**********************************************************************/
99 close_safely_handler (int signo)
101 EMACS_REESTABLISH_SIGNAL (signo, close_safely_handler);
106 close_safely (int fd)
109 signal (SIGALRM, close_safely_handler);
117 close_descriptor_pair (int in, int out)
121 if (out != in && out >= 0)
125 /* Close all descriptors currently in use for communication
126 with subprocess. This is used in a newly-forked subprocess
127 to get rid of irrelevant descriptors. */
130 close_process_descs_mapfun (const void* key, void* contents, void* arg)
133 CVOID_TO_LISP (proc, contents);
134 event_stream_delete_stream_pair (XPROCESS(proc)->pipe_instream,
135 XPROCESS(proc)->pipe_outstream);
139 /* #### This function is currently called from child_setup
140 in callproc.c. It should become static though - kkm */
142 close_process_descs (void)
144 maphash (close_process_descs_mapfun, usid_to_process, 0);
147 /* connect to an existing file descriptor. This is very similar to
148 open-network-stream except that it assumes that the connection has
149 already been initialized. It is currently used for ToolTalk
152 /* This function used to be visible on the Lisp level, but there is no
153 real point in doing that. Here is the doc string:
155 "Connect to an existing file descriptor.\n\
156 Returns a subprocess-object to represent the connection.\n\
157 Input and output work as for subprocesses; `delete-process' closes it.\n\
158 Args are NAME BUFFER INFD OUTFD.\n\
159 NAME is name for process. It is modified if necessary to make it unique.\n\
160 BUFFER is the buffer (or buffer-name) to associate with the process.\n\
161 Process output goes at end of that buffer, unless you specify\n\
162 an output stream or filter function to handle the output.\n\
163 BUFFER may be also nil, meaning that this process is not associated\n\
165 INFD and OUTFD specify the file descriptors to use for input and\n\
166 output, respectively."
170 connect_to_file_descriptor (Lisp_Object name, Lisp_Object buffer,
171 Lisp_Object infd, Lisp_Object outfd)
173 /* This function can GC */
182 if (get_process_from_usid (FD_TO_USID(inch)))
183 error ("There is already a process connected to fd %d", inch);
185 buffer = Fget_buffer_create (buffer);
186 proc = make_process_internal (name);
188 XPROCESS (proc)->pid = Fcons (infd, name);
189 XPROCESS (proc)->buffer = buffer;
190 init_process_io_handles (XPROCESS (proc), (void*)inch, (void*)XINT (outfd), 0);
191 UNIX_DATA (XPROCESS (proc))->connected_via_filedesc_p = 1;
193 event_stream_select_process (XPROCESS (proc));
200 /* Open an available pty, returning a file descriptor.
201 Return -1 on failure.
202 The file name of the terminal corresponding to the pty
203 is left in the variable pty_name. */
211 /* Some systems name their pseudoterminals so that there are gaps in
212 the usual sequence - for example, on HP9000/S700 systems, there
213 are no pseudoterminals with names ending in 'f'. So we wait for
214 three failures in a row before deciding that we've reached the
216 int failed_count = 0;
227 for (c = FIRST_PTY_LETTER; c <= 'z'; c++)
228 for (i = 0; i < 16; i++)
231 #ifdef PTY_NAME_SPRINTF
234 sprintf (pty_name, "/dev/pty%c%x", c, i);
235 #endif /* no PTY_NAME_SPRINTF */
239 #else /* no PTY_OPEN */
241 /* Unusual IRIS code */
242 *ptyv = open ("/dev/ptc", O_RDWR | O_NONBLOCK | OPEN_BINARY, 0);
245 if (fstat (fd, &stb) < 0)
248 if (stat (pty_name, &stb) < 0)
251 if (failed_count >= 3)
256 fd = open (pty_name, O_RDWR | O_NONBLOCK | OPEN_BINARY, 0);
257 #endif /* not IRIS */
258 #endif /* no PTY_OPEN */
262 /* check to make certain that both sides are available
263 this avoids a nasty yet stupid bug in rlogins */
264 #ifdef PTY_TTY_NAME_SPRINTF
267 sprintf (pty_name, "/dev/tty%c%x", c, i);
268 #endif /* no PTY_TTY_NAME_SPRINTF */
269 #if !defined(UNIPLUS) && !defined(HAVE_GETPT)
270 if (access (pty_name, 6) != 0)
273 #if !defined(IRIS) && !defined(__sgi)
279 #endif /* not UNIPLUS */
286 #endif /* HAVE_PTYS */
289 create_bidirectional_pipe (int *inchannel, int *outchannel,
290 volatile int *forkin, volatile int *forkout)
295 if (socketpair (AF_UNIX, SOCK_STREAM, 0, sv) < 0)
297 *outchannel = *inchannel = sv[0];
298 *forkout = *forkin = sv[1];
299 #else /* not SKTPAIR */
302 if (temp < 0) return -1;
306 if (temp < 0) return -1;
309 #endif /* not SKTPAIR */
316 #if !(defined(HAVE_GETADDRINFO) && defined(HAVE_GETNAMEINFO))
318 get_internet_address (Lisp_Object host, struct sockaddr_in *address,
321 struct hostent *host_info_ptr = NULL;
331 if (count++ > 10) break;
332 #ifndef BROKEN_CYGWIN
336 /* Some systems can't handle SIGIO/SIGALARM in gethostbyname. */
337 slow_down_interrupts ();
338 host_info_ptr = gethostbyname ((char *) XSTRING_DATA (host));
339 speed_up_interrupts ();
341 if (! (host_info_ptr == 0 && h_errno == TRY_AGAIN))
344 Fsleep_for (make_int (1));
348 address->sin_family = host_info_ptr->h_addrtype;
349 memcpy (&address->sin_addr, host_info_ptr->h_addr, host_info_ptr->h_length);
353 IN_ADDR numeric_addr;
354 /* Attempt to interpret host as numeric inet address */
355 numeric_addr = inet_addr ((char *) XSTRING_DATA (host));
356 if (NUMERIC_ADDR_ERROR)
358 maybe_error (Qprocess, errb,
359 "Unknown host \"%s\"", XSTRING_DATA (host));
363 /* There was some broken code here that called strlen() here
364 on (char *) &numeric_addr and even sometimes accessed
365 uninitialized data. */
366 address->sin_family = AF_INET;
367 * (IN_ADDR *) &address->sin_addr = numeric_addr;
372 #endif /* !(HAVE_GETADDRINFO && HAVE_GETNAMEINFO) */
375 set_socket_nonblocking_maybe (int fd, int port, const char* proto)
377 #ifdef PROCESS_IO_BLOCKING
380 for (tail = network_stream_blocking_port_list; CONSP (tail); tail = XCDR (tail))
382 Lisp_Object tail_port = XCAR (tail);
384 if (STRINGP (tail_port))
386 struct servent *svc_info;
387 CHECK_STRING (tail_port);
388 svc_info = getservbyname ((char *) XSTRING_DATA (tail_port), proto);
389 if ((svc_info != 0) && (svc_info->s_port == port))
394 else if (INTP (tail_port) && (htons ((unsigned short) XINT (tail_port)) == port))
400 set_descriptor_non_blocking (fd);
403 set_descriptor_non_blocking (fd);
404 #endif /* PROCESS_IO_BLOCKING */
407 #endif /* HAVE_SOCKETS */
409 /* Compute the Lisp form of the process status from
410 the numeric status that was returned by `wait'. */
413 update_status_from_wait_code (Lisp_Process *p, int *w_fmh)
415 /* C compiler lossage when attempting to pass w directly */
420 p->status_symbol = Qstop;
421 p->exit_code = WSTOPSIG (w);
424 else if (WIFEXITED (w))
426 p->status_symbol = Qexit;
427 p->exit_code = WEXITSTATUS (w);
430 else if (WIFSIGNALED (w))
432 p->status_symbol = Qsignal;
433 p->exit_code = WTERMSIG (w);
434 p->core_dumped = WCOREDUMP (w);
438 p->status_symbol = Qrun;
445 #define MAX_EXITED_PROCESSES 1000
446 static volatile pid_t exited_processes[MAX_EXITED_PROCESSES];
447 static volatile int exited_processes_status[MAX_EXITED_PROCESSES];
448 static volatile int exited_processes_index;
450 static volatile int sigchld_happened;
452 /* On receipt of a signal that a child status has changed,
453 loop asking about children with changed statuses until
454 the system says there are no more. All we do is record
455 the processes and wait status.
457 This function could be called from within the SIGCHLD
458 handler, so it must be completely reentrant. When
459 not called from a SIGCHLD handler, BLOCK_SIGCHLD should
460 be non-zero so that SIGCHLD is blocked while this
461 function is running. (This is necessary so avoid
462 race conditions with the SIGCHLD_HAPPENED flag). */
465 record_exited_processes (int block_sigchld)
467 if (!sigchld_happened)
472 #ifdef EMACS_BLOCK_SIGNAL
474 EMACS_BLOCK_SIGNAL (SIGCHLD);
477 while (sigchld_happened)
482 /* Keep trying to get a status until we get a definitive result. */
489 # endif /* not WUNTRACED */
491 pid = waitpid ((pid_t) -1, &w, WNOHANG | WUNTRACED);
493 pid = wait3 (&w, WNOHANG | WUNTRACED, 0);
495 #else /* not WNOHANG */
497 #endif /* not WNOHANG */
499 while (pid <= 0 && errno == EINTR);
504 if (exited_processes_index < MAX_EXITED_PROCESSES)
506 exited_processes[exited_processes_index] = pid;
507 exited_processes_status[exited_processes_index] = w;
508 exited_processes_index++;
511 /* On systems with WNOHANG, we just ignore the number
512 of times that SIGCHLD was signalled, and keep looping
513 until there are no more processes to wait on. If we
514 don't have WNOHANG, we have to rely on the count in
518 #endif /* not WNOHANG */
521 sigchld_happened = 0;
524 EMACS_UNBLOCK_SIGNAL (SIGCHLD);
527 /* For any processes that have changed status and are recorded
528 and such, update the corresponding Lisp_Process.
529 We separate this from record_exited_processes() so that
530 we never have to call this function from within a signal
531 handler. We block SIGCHLD in case record_exited_processes()
532 is called from a signal handler. */
534 /** USG WARNING: Although it is not obvious from the documentation
535 in signal(2), on a USG system the SIGCLD handler MUST NOT call
536 signal() before executing at least one wait(), otherwise the handler
537 will be called again, resulting in an infinite loop. The relevant
538 portion of the documentation reads "SIGCLD signals will be queued
539 and the signal-catching function will be continually reentered until
540 the queue is empty". Invoking signal() causes the kernel to reexamine
541 the SIGCLD queue. Fred Fish, UniSoft Systems Inc.
543 (Note that now this only applies in SYS V Release 2 and before.
544 On SYS V Release 3, we use sigset() to set the signal handler for
545 the first time, and so we don't have to reestablish the signal handler
546 in the handler below. On SYS V Release 4, we don't get this weirdo
547 behavior when we use sigaction(), which we do use.) */
550 sigchld_handler (int signo)
552 #ifdef OBNOXIOUS_SYSV_SIGCLD_BEHAVIOR
553 int old_errno = errno;
556 record_exited_processes (0);
561 #ifdef HAVE_UNIXOID_EVENT_LOOP
562 signal_fake_event ();
564 /* WARNING - must come after wait3() for USG systems */
565 EMACS_REESTABLISH_SIGNAL (signo, sigchld_handler);
571 #ifdef SIGNALS_VIA_CHARACTERS
572 /* Get signal character to send to process if SIGNALS_VIA_CHARACTERS */
575 process_signal_char (int tty_fd, int signo)
577 /* If it's not a tty, pray that these default values work */
578 if (!isatty(tty_fd)) {
579 #define CNTL(ch) (037 & (ch))
582 case SIGINT: return CNTL('C');
583 case SIGQUIT: return CNTL('\\');
585 case SIGTSTP: return CNTL('Z');
591 /* TERMIOS is the latest and bestest, and seems most likely to work.
592 If the system has it, use it. */
595 tcgetattr (tty_fd, &t);
598 case SIGINT: return t.c_cc[VINTR];
599 case SIGQUIT: return t.c_cc[VQUIT];
600 #if defined(SIGTSTP) && defined(VSUSP)
601 case SIGTSTP: return t.c_cc[VSUSP];
606 # elif defined (TIOCGLTC) && defined (TIOCGETC) /* not HAVE_TERMIOS */
608 /* On Berkeley descendants, the following IOCTL's retrieve the
609 current control characters. */
614 case SIGINT: ioctl (tty_fd, TIOCGETC, &c); return c.t_intrc;
615 case SIGQUIT: ioctl (tty_fd, TIOCGETC, &c); return c.t_quitc;
617 case SIGTSTP: ioctl (tty_fd, TIOCGLTC, &lc); return lc.t_suspc;
618 # endif /* SIGTSTP */
622 # elif defined (TCGETA) /* ! defined (TIOCGLTC) && defined (TIOCGETC) */
624 /* On SYSV descendants, the TCGETA ioctl retrieves the current
625 control characters. */
627 ioctl (tty_fd, TCGETA, &t);
629 case SIGINT: return t.c_cc[VINTR];
630 case SIGQUIT: return t.c_cc[VQUIT];
632 case SIGTSTP: return t.c_cc[VSWTCH];
633 # endif /* SIGTSTP */
636 # else /* ! defined (TCGETA) */
637 #error ERROR! Using SIGNALS_VIA_CHARACTERS, but not HAVE_TERMIOS || (TIOCGLTC && TIOCGETC) || TCGETA
638 /* If your system configuration files define SIGNALS_VIA_CHARACTERS,
639 you'd better be using one of the alternatives above! */
640 # endif /* ! defined (TCGETA) */
643 #endif /* SIGNALS_VIA_CHARACTERS */
648 /**********************************************************************/
649 /* Process implementation methods */
650 /**********************************************************************/
653 * Allocate and initialize Lisp_Process->process_data
657 unix_alloc_process_data (Lisp_Process *p)
659 p->process_data = xnew (struct unix_process_data);
661 UNIX_DATA(p)->connected_via_filedesc_p = 0;
662 UNIX_DATA(p)->infd = -1;
663 UNIX_DATA(p)->subtty = -1;
664 UNIX_DATA(p)->tty_name = Qnil;
665 UNIX_DATA(p)->pty_flag = 0;
669 * Mark any Lisp objects in Lisp_Process->process_data
673 unix_mark_process_data (Lisp_Process *proc)
675 mark_object (UNIX_DATA(proc)->tty_name);
679 * Initialize XEmacs process implementation once
684 unix_init_process (void)
687 if (! noninteractive || initialized)
689 signal (SIGCHLD, sigchld_handler);
694 * Initialize any process local data. This is called when newly
695 * created process is connected to real OS file handles. The
696 * handles are generally represented by void* type, but are
697 * of type int (file descriptors) for UNIX
701 unix_init_process_io_handles (Lisp_Process *p, void* in, void* out, int flags)
703 UNIX_DATA(p)->infd = (int)in;
707 * Fork off a subprocess. P is a pointer to a newly created subprocess
708 * object. If this function signals, the caller is responsible for
709 * deleting (and finalizing) the process object.
711 * The method must return PID of the new process, a (positive??? ####) number
712 * which fits into Lisp_Int. No return value indicates an error, the method
713 * must signal an error instead.
717 unix_create_process (Lisp_Process *p,
718 Lisp_Object *argv, int nargv,
719 Lisp_Object program, Lisp_Object cur_dir)
721 /* This function rewritten by ben@xemacs.org. */
726 /* Use volatile to protect variables from being clobbered by longjmp. */
727 volatile int forkin = -1;
728 volatile int forkout = -1;
729 volatile int pty_flag = 0;
732 if (!NILP (Vprocess_connection_type))
734 /* find a new pty, open the master side, return the opened
735 file handle, and store the name of the corresponding slave
736 side in global variable pty_name. */
737 outchannel = inchannel = allocate_pty ();
742 /* You're "supposed" to now open the slave in the child.
743 On some systems, we can open it here; this allows for
744 better error checking. */
746 /* On USG systems it does not work to open the pty's tty here
747 and then close and reopen it in the child. */
749 /* Don't let this terminal become our controlling terminal
750 (in case we don't have one). */
751 forkout = forkin = open (pty_name, O_RDWR | O_NOCTTY | OPEN_BINARY, 0);
753 forkout = forkin = open (pty_name, O_RDWR | OPEN_BINARY, 0);
758 UNIX_DATA(p)->pty_flag = pty_flag = 1;
761 #endif /* HAVE_PTYS */
762 if (create_bidirectional_pipe (&inchannel, &outchannel,
763 &forkin, &forkout) < 0)
767 /* Replaced by close_process_descs */
768 set_exclusive_use (inchannel);
769 set_exclusive_use (outchannel);
772 set_descriptor_non_blocking (inchannel);
774 /* Record this as an active process, with its channels.
775 As a result, child_setup will close Emacs's side of the pipes. */
776 init_process_io_handles (p, (void*)inchannel, (void*)outchannel,
777 pty_flag ? STREAM_PTY_FLUSHING : 0);
778 /* Record the tty descriptor used in the subprocess. */
779 UNIX_DATA(p)->subtty = forkin;
782 #if !defined(__CYGWIN32__)
783 /* child_setup must clobber environ on systems with true vfork.
784 Protect it from permanent change. */
785 char **save_environ = environ;
791 /**** Now we're in the child process ****/
792 int xforkin = forkin;
793 int xforkout = forkout;
796 EMACS_SEPARATE_PROCESS_GROUP ();
800 /* Disconnect the current controlling terminal, pursuant to
801 making the pty be the controlling terminal of the process.
802 Also put us in our own process group. */
804 disconnect_controlling_terminal ();
806 /* Open the pty connection and make the pty's terminal
807 our controlling terminal.
809 On systems with TIOCSCTTY, we just use it to set
810 the controlling terminal. On other systems, the
811 first TTY we open becomes the controlling terminal.
812 So, we end up with four possibilities:
814 (1) on USG and TIOCSCTTY systems, we open the pty
816 (2) on other USG systems, we just open the pty.
817 (3) on non-USG systems with TIOCSCTTY, we
818 just use TIOCSCTTY. (On non-USG systems, we
819 already opened the pty in the parent process.)
820 (4) on non-USG systems without TIOCSCTTY, we
821 close the pty and reopen it.
823 This would be cleaner if we didn't open the pty
824 in the parent process, but doing it that way
825 makes it possible to trap error conditions.
826 It's harder to convey an error from the child
827 process, and I don't feel like messing with
830 /* There was some weirdo, probably wrong,
831 conditionalization on RTU and UNIPLUS here.
832 I deleted it. So sue me. */
834 /* SunOS has TIOCSCTTY but the close/open method
837 # if defined (USG) || !defined (TIOCSCTTY)
838 /* Now close the pty (if we had it open) and reopen it.
839 This makes the pty the controlling terminal of the
841 /* I wonder if close (open (pty_name, ...)) would work? */
844 xforkout = xforkin = open (pty_name, O_RDWR | OPEN_BINARY, 0);
847 write (1, "Couldn't open the pty terminal ", 31);
848 write (1, pty_name, strlen (pty_name));
852 # endif /* USG or not TIOCSCTTY */
854 /* Miscellaneous setup required for some systems.
855 Must be done before using tc* functions on xforkin.
856 This guarantees that isatty(xforkin) is true. */
858 # ifdef SETUP_SLAVE_PTY
860 # endif /* SETUP_SLAVE_PTY */
863 /* We ignore the return value
864 because faith@cs.unc.edu says that is necessary on Linux. */
865 assert (isatty (xforkin));
866 ioctl (xforkin, TIOCSCTTY, 0);
867 # endif /* TIOCSCTTY */
869 /* Change the line discipline. */
871 # if defined (HAVE_TERMIOS) && defined (LDISC1)
874 assert (isatty (xforkin));
875 tcgetattr (xforkin, &t);
877 if (tcsetattr (xforkin, TCSANOW, &t) < 0)
878 perror ("create_process/tcsetattr LDISC1 failed\n");
880 # elif defined (NTTYDISC) && defined (TIOCSETD)
882 /* Use new line discipline. TIOCSETD is accepted and
883 ignored on Sys5.4 systems with ttcompat. */
884 int ldisc = NTTYDISC;
885 assert (isatty (xforkin));
886 ioctl (xforkin, TIOCSETD, &ldisc);
888 # endif /* TIOCSETD & NTTYDISC */
890 /* Make our process group be the foreground group
891 of our new controlling terminal. */
894 int piddly = EMACS_GET_PROCESS_GROUP ();
895 EMACS_SET_TTY_PROCESS_GROUP (xforkin, &piddly);
898 /* On AIX, we've disabled SIGHUP above once we start a
899 child on a pty. Now reenable it in the child, so it
900 will die when we want it to.
901 JV: This needs to be done ALWAYS as we might have inherited
902 a SIG_IGN handling from our parent (nohup) and we are in new
905 signal (SIGHUP, SIG_DFL);
909 /* Set up the terminal characteristics of the pty. */
910 child_setup_tty (xforkout);
912 #endif /* HAVE_PTYS */
914 signal (SIGINT, SIG_DFL);
915 signal (SIGQUIT, SIG_DFL);
919 char **new_argv = alloca_array (char *, nargv + 2);
922 /* Nothing below here GCs so our string pointers shouldn't move. */
923 new_argv[0] = (char *) XSTRING_DATA (program);
924 for (i = 0; i < nargv; i++)
926 CHECK_STRING (argv[i]);
927 new_argv[i + 1] = (char *) XSTRING_DATA (argv[i]);
931 TO_EXTERNAL_FORMAT (LISP_STRING, cur_dir,
932 C_STRING_ALLOCA, current_dir,
935 child_setup (xforkin, xforkout, xforkout, new_argv, current_dir);
938 } /**** End of child code ****/
940 /**** Back in parent process ****/
941 #if !defined(__CYGWIN32__)
942 environ = save_environ;
948 close_descriptor_pair (forkin, forkout);
949 report_file_error ("Doing fork", Qnil);
952 /* #### dmoore - why is this commented out, otherwise we leave
953 subtty = forkin, but then we close forkin just below. */
954 /* UNIX_DATA(p)->subtty = -1; */
956 /* If the subfork execv fails, and it exits,
957 this close hangs. I don't know why.
958 So have an interrupt jar it loose. */
960 close_safely (forkin);
961 if (forkin != forkout && forkout >= 0)
966 UNIX_DATA (p)->tty_name = build_string (pty_name);
969 UNIX_DATA (p)->tty_name = Qnil;
971 /* Notice that SIGCHLD was not blocked. (This is not possible on
972 some systems.) No biggie if SIGCHLD occurs right around the
973 time that this call happens, because SIGCHLD() does not actually
974 deselect the process (that doesn't occur until the next time
975 we're waiting for an event, when status_notify() is called). */
980 int save_errno = errno;
981 close_descriptor_pair (forkin, forkout);
982 close_descriptor_pair (inchannel, outchannel);
984 report_file_error ("Opening pty or pipe", Qnil);
985 return 0; /* not reached */
989 /* Return nonzero if this process is a ToolTalk connection. */
992 unix_tooltalk_connection_p (Lisp_Process *p)
994 return UNIX_DATA(p)->connected_via_filedesc_p;
997 /* This is called to set process' virtual terminal size */
1000 unix_set_window_size (Lisp_Process* p, int cols, int rows)
1002 return set_window_size (UNIX_DATA(p)->infd, cols, rows);
1006 * This method is called to update status fields of the process
1007 * structure. If the process has not existed, this method is
1008 * expected to do nothing.
1010 * The method is called only for real child processes.
1015 unix_update_status_if_terminated (Lisp_Process* p)
1019 EMACS_BLOCK_SIGNAL (SIGCHLD);
1021 if (waitpid (XINT (p->pid), &w, WNOHANG) == XINT (p->pid))
1024 update_status_from_wait_code (p, &w);
1027 EMACS_UNBLOCK_SIGNAL (SIGCHLD);
1033 * Update status of all exited processes. Called when SIGCLD has signaled.
1038 unix_reap_exited_processes (void)
1043 #ifndef OBNOXIOUS_SYSV_SIGCLD_BEHAVIOR
1044 record_exited_processes (1);
1047 if (exited_processes_index <= 0)
1052 #ifdef EMACS_BLOCK_SIGNAL
1053 EMACS_BLOCK_SIGNAL (SIGCHLD);
1055 for (i = 0; i < exited_processes_index; i++)
1057 int pid = exited_processes[i];
1058 int w = exited_processes_status[i];
1060 /* Find the process that signaled us, and record its status. */
1065 LIST_LOOP (tail, Vprocess_list)
1067 Lisp_Object proc = XCAR (tail);
1068 p = XPROCESS (proc);
1069 if (INTP (p->pid) && XINT (p->pid) == pid)
1077 /* Change the status of the process that was found. */
1080 update_status_from_wait_code (p, &w);
1082 /* If process has terminated, stop waiting for its output. */
1083 if (WIFSIGNALED (w) || WIFEXITED (w))
1085 if (!NILP(p->pipe_instream))
1087 /* We can't just call event_stream->unselect_process_cb (p)
1088 here, because that calls XtRemoveInput, which is not
1089 necessarily reentrant, so we can't call this at interrupt
1097 /* There was no asynchronous process found for that id. Check
1098 if we have a synchronous process. Only set sync process status
1099 if there is one, so we work OK with the waitpid() call in
1100 wait_for_termination(). */
1101 if (synch_process_alive != 0)
1102 { /* Set the global sync process status variables. */
1103 synch_process_alive = 0;
1105 /* Report the status of the synchronous process. */
1107 synch_process_retcode = WEXITSTATUS (w);
1108 else if (WIFSIGNALED (w))
1109 synch_process_death = signal_name (WTERMSIG (w));
1114 exited_processes_index = 0;
1116 EMACS_UNBLOCK_SIGNAL (SIGCHLD);
1118 #endif /* SIGCHLD */
1121 * Stuff the entire contents of LSTREAM to the process output pipe
1124 static JMP_BUF send_process_frame;
1127 send_process_trap (int signum)
1129 EMACS_REESTABLISH_SIGNAL (signum, send_process_trap);
1130 EMACS_UNBLOCK_SIGNAL (signum);
1131 LONGJMP (send_process_frame, 1);
1135 unix_send_process (Lisp_Object proc, struct lstream* lstream)
1137 /* Use volatile to protect variables from being clobbered by longjmp. */
1138 SIGTYPE (*volatile old_sigpipe) (int) = 0;
1139 volatile Lisp_Object vol_proc = proc;
1140 Lisp_Process *volatile p = XPROCESS (proc);
1142 /* #### JV: layering violation?
1144 This function knows too much about the relation between the encodingstream
1145 (DATA_OUTSTREAM) and te actual output stream p->output_stream.
1147 If encoding streams properly forwarded all calls, we could simply
1148 use DATA_OUTSTREAM everywhere. */
1150 if (!SETJMP (send_process_frame))
1152 /* use a reasonable-sized buffer (somewhere around the size of the
1153 stream buffer) so as to avoid inundating the stream with blocked
1155 Bufbyte chunkbuf[512];
1162 chunklen = Lstream_read (lstream, chunkbuf, 512);
1164 break; /* perhaps should abort() if < 0?
1165 This should never happen. */
1167 (SIGTYPE (*) (int)) signal (SIGPIPE, send_process_trap);
1168 /* Lstream_write() will never successfully write less than
1169 the amount sent in. In the worst case, it just buffers
1170 the unwritten data. */
1171 writeret = Lstream_write (XLSTREAM (DATA_OUTSTREAM(p)), chunkbuf,
1173 signal (SIGPIPE, old_sigpipe);
1175 /* This is a real error. Blocking errors are handled
1176 specially inside of the filedesc stream. */
1177 report_file_error ("writing to process", list1 (proc));
1178 while (Lstream_was_blocked_p (XLSTREAM (p->pipe_outstream)))
1180 /* Buffer is full. Wait, accepting input;
1181 that may allow the program
1182 to finish doing output and read more. */
1183 Faccept_process_output (Qnil, make_int (1), Qnil);
1184 /* It could have *really* finished, deleting the process */
1185 if (NILP(p->pipe_outstream))
1188 (SIGTYPE (*) (int)) signal (SIGPIPE, send_process_trap);
1189 Lstream_flush (XLSTREAM (p->pipe_outstream));
1190 signal (SIGPIPE, old_sigpipe);
1195 { /* We got here from a longjmp() from the SIGPIPE handler */
1196 signal (SIGPIPE, old_sigpipe);
1197 /* Close the file lstream so we don't attempt to write to it further */
1198 /* #### There is controversy over whether this might cause fd leakage */
1199 /* my tests say no. -slb */
1200 XLSTREAM (p->pipe_outstream)->flags &= ~LSTREAM_FL_IS_OPEN;
1201 p->status_symbol = Qexit;
1202 p->exit_code = 256; /* #### SIGPIPE ??? */
1206 deactivate_process (*((Lisp_Object *) (&vol_proc)));
1207 error ("SIGPIPE raised on process %s; closed it",
1208 XSTRING_DATA (p->name));
1211 old_sigpipe = (SIGTYPE (*) (int)) signal (SIGPIPE, send_process_trap);
1212 Lstream_flush (XLSTREAM (DATA_OUTSTREAM(p)));
1213 signal (SIGPIPE, old_sigpipe);
1217 * Send EOF to the process. The default implementation simply
1218 * closes the output stream. The method must return 0 to call
1219 * the default implementation, or 1 if it has taken all care about
1220 * sending EOF to the process.
1224 unix_process_send_eof (Lisp_Object proc)
1226 if (!UNIX_DATA (XPROCESS (proc))->pty_flag)
1229 /* #### get_eof_char simply doesn't return the correct character
1230 here. Maybe it is needed to determine the right eof
1231 character in init_process_io_handles but here it simply screws
1234 Bufbyte eof_char = get_eof_char (XPROCESS (proc));
1235 send_process (proc, Qnil, &eof_char, 0, 1);
1237 send_process (proc, Qnil, (const Bufbyte *) "\004", 0, 1);
1243 * Called before the process is deactivated. The process object
1244 * is not immediately finalized, just undergoes a transition to
1247 * The return value is a unique stream ID, as returned by
1248 * event_stream_delete_stream_pair
1250 * In the lack of this method, only event_stream_delete_stream_pair
1251 * is called on both I/O streams of the process.
1253 * The UNIX version guards this by ignoring possible SIGPIPE.
1257 unix_deactivate_process (Lisp_Process *p)
1259 SIGTYPE (*old_sigpipe) (int) = 0;
1262 if (UNIX_DATA(p)->infd >= 0)
1263 flush_pending_output (UNIX_DATA(p)->infd);
1265 /* closing the outstream could result in SIGPIPE, so ignore it. */
1266 old_sigpipe = (SIGTYPE (*) (int)) signal (SIGPIPE, SIG_IGN);
1267 usid = event_stream_delete_stream_pair (p->pipe_instream, p->pipe_outstream);
1268 signal (SIGPIPE, old_sigpipe);
1270 UNIX_DATA(p)->infd = -1;
1275 /* send a signal number SIGNO to PROCESS.
1276 CURRENT_GROUP means send to the process group that currently owns
1277 the terminal being used to communicate with PROCESS.
1278 This is used for various commands in shell mode.
1279 If NOMSG is zero, insert signal-announcements into process's buffers
1282 If we can, we try to signal PROCESS by sending control characters
1283 down the pty. This allows us to signal inferiors who have changed
1284 their uid, for which killpg would return an EPERM error.
1286 The method signals an error if the given SIGNO is not valid
1290 unix_kill_child_process (Lisp_Object proc, int signo,
1291 int current_group, int nomsg)
1296 Lisp_Process *p = XPROCESS (proc);
1298 if (!UNIX_DATA(p)->pty_flag)
1301 /* If we are using pgrps, get a pgrp number and make it negative. */
1304 #ifdef SIGNALS_VIA_CHARACTERS
1305 /* If possible, send signals to the entire pgrp
1306 by sending an input character to it. */
1308 char sigchar = process_signal_char(UNIX_DATA(p)->subtty, signo);
1310 send_process (proc, Qnil, (Bufbyte *) &sigchar, 0, 1);
1314 #endif /* ! defined (SIGNALS_VIA_CHARACTERS) */
1317 /* Get the pgrp using the tty itself, if we have that.
1318 Otherwise, use the pty to get the pgrp.
1319 On pfa systems, saka@pfu.fujitsu.co.JP writes:
1320 "TIOCGPGRP symbol defined in sys/ioctl.h at E50.
1321 But, TIOCGPGRP does not work on E50 ;-P works fine on E60"
1322 His patch indicates that if TIOCGPGRP returns an error, then
1323 we should just assume that p->pid is also the process group id. */
1327 err = ioctl ( (UNIX_DATA(p)->subtty != -1
1328 ? UNIX_DATA(p)->subtty
1329 : UNIX_DATA(p)->infd), TIOCGPGRP, &gid);
1333 gid = - XINT (p->pid);
1334 #endif /* ! defined (pfa) */
1340 #else /* ! defined (TIOCGPGRP ) */
1341 /* Can't select pgrps on this system, so we know that
1342 the child itself heads the pgrp. */
1343 gid = - XINT (p->pid);
1344 #endif /* ! defined (TIOCGPGRP ) */
1347 gid = - XINT (p->pid);
1353 p->status_symbol = Qrun;
1360 #endif /* ! defined (SIGCONT) */
1364 flush_pending_output (UNIX_DATA(p)->infd);
1368 /* If we don't have process groups, send the signal to the immediate
1369 subprocess. That isn't really right, but it's better than any
1370 obvious alternative. */
1373 kill_retval = kill (XINT (p->pid), signo) ? errno : 0;
1377 /* gid may be a pid, or minus a pgrp's number */
1378 #if defined (TIOCSIGNAL) || defined (TIOCSIGSEND)
1382 kill_retval = ioctl (UNIX_DATA(p)->infd, TIOCSIGNAL, signo);
1383 #else /* ! defined (TIOCSIGNAL) */
1384 kill_retval = ioctl (UNIX_DATA(p)->infd, TIOCSIGSEND, signo);
1385 #endif /* ! defined (TIOCSIGNAL) */
1388 kill_retval = kill (- XINT (p->pid), signo) ? errno : 0;
1389 #else /* ! (defined (TIOCSIGNAL) || defined (TIOCSIGSEND)) */
1390 kill_retval = EMACS_KILLPG (-gid, signo) ? errno : 0;
1391 #endif /* ! (defined (TIOCSIGNAL) || defined (TIOCSIGSEND)) */
1394 if (kill_retval < 0 && errno == EINVAL)
1395 error ("Signal number %d is invalid for this system", signo);
1399 * Kill any process in the system given its PID.
1401 * Returns zero if a signal successfully sent, or
1402 * negative number upon failure
1406 unix_kill_process_by_pid (int pid, int sigcode)
1408 return kill (pid, sigcode);
1412 * Return TTY name used to communicate with subprocess
1416 unix_get_tty_name (Lisp_Process *p)
1418 return UNIX_DATA (p)->tty_name;
1422 * Canonicalize host name HOST, and return its canonical form
1424 * The default implementation just takes HOST for a canonical name.
1429 unix_canonicalize_host_name (Lisp_Object host)
1431 #if defined(HAVE_GETADDRINFO) && defined(HAVE_GETNAMEINFO)
1432 struct addrinfo hints, *res;
1433 static char addrbuf[NI_MAXHOST];
1434 Lisp_Object canonname;
1439 hints.ai_flags = AI_CANONNAME;
1440 hints.ai_family = AF_UNSPEC;
1441 hints.ai_socktype = SOCK_STREAM;
1442 hints.ai_protocol = 0;
1443 TO_EXTERNAL_FORMAT (LISP_STRING, host, C_STRING_ALLOCA, ext_host, Qnative);
1444 retval = getaddrinfo (ext_host, NULL, &hints, &res);
1449 TO_INTERNAL_FORMAT (C_STRING, gai_strerror (retval),
1450 C_STRING_ALLOCA, gai_error,
1452 maybe_error (Qprocess, ERROR_ME_NOT,
1453 "%s \"%s\"", gai_error, XSTRING_DATA (host));
1458 int gni = getnameinfo (res->ai_addr, res->ai_addrlen,
1459 addrbuf, sizeof(addrbuf),
1460 NULL, 0, NI_NUMERICHOST);
1461 canonname = gni ? host : build_ext_string (addrbuf, Qnative);
1467 #else /* ! HAVE_GETADDRINFO */
1468 struct sockaddr_in address;
1470 if (!get_internet_address (host, &address, ERROR_ME_NOT))
1473 if (address.sin_family == AF_INET)
1474 return build_string (inet_ntoa (address.sin_addr));
1476 /* #### any clue what to do here? */
1478 #endif /* ! HAVE_GETADDRINFO */
1481 /* open a TCP network connection to a given HOST/SERVICE. Treated
1482 exactly like a normal process when reading and writing. Only
1483 differences are in status display and process deletion. A network
1484 connection has no PID; you cannot signal it. All you can do is
1485 deactivate and close it via delete-process */
1488 unix_open_network_stream (Lisp_Object name, Lisp_Object host, Lisp_Object service,
1489 Lisp_Object protocol, void** vinfd, void** voutfd)
1495 volatile int retry = 0;
1498 CHECK_STRING (host);
1500 if (!EQ (protocol, Qtcp) && !EQ (protocol, Qudp))
1501 error ("Unsupported protocol \"%s\"",
1502 string_data (symbol_name (XSYMBOL (protocol))));
1505 #if defined(HAVE_GETADDRINFO) && defined(HAVE_GETNAMEINFO)
1506 struct addrinfo hints, *res;
1507 struct addrinfo * volatile lres;
1509 volatile int xerrno = 0;
1510 volatile int failed_connect = 0;
1513 * Caution: service can either be a string or int.
1514 * Convert to a C string for later use by getaddrinfo.
1519 snprintf (portbuf, sizeof (portbuf), "%ld", (long) XINT (service));
1520 portstring = portbuf;
1521 port = htons ((unsigned short) XINT (service));
1525 CHECK_STRING (service);
1526 TO_EXTERNAL_FORMAT (LISP_STRING, service,
1527 C_STRING_ALLOCA, portstring,
1534 hints.ai_family = AF_UNSPEC;
1535 if (EQ (protocol, Qtcp))
1536 hints.ai_socktype = SOCK_STREAM;
1537 else /* EQ (protocol, Qudp) */
1538 hints.ai_socktype = SOCK_DGRAM;
1539 hints.ai_protocol = 0;
1540 TO_EXTERNAL_FORMAT (LISP_STRING, host, C_STRING_ALLOCA, ext_host, Qnative);
1541 retval = getaddrinfo (ext_host, portstring, &hints, &res);
1546 TO_INTERNAL_FORMAT (C_STRING, gai_strerror (retval),
1547 C_STRING_ALLOCA, gai_error,
1549 error ("%s/%s %s", XSTRING_DATA (host), portstring, gai_error);
1553 for (lres = res; lres ; lres = lres->ai_next)
1555 if (EQ (protocol, Qtcp))
1556 s = socket (lres->ai_family, SOCK_STREAM, 0);
1557 else /* EQ (protocol, Qudp) */
1558 s = socket (lres->ai_family, SOCK_DGRAM, 0);
1563 /* Turn off interrupts here -- see comments below. There used to
1564 be code which called bind_polling_period() to slow the polling
1565 period down rather than turn it off, but that seems rather
1566 bogus to me. Best thing here is to use a non-blocking connect
1567 or something, to check for QUIT. */
1569 /* Comments that are not quite valid: */
1571 /* Kernel bugs (on Ultrix at least) cause lossage (not just EINTR)
1572 when connect is interrupted. So let's not let it get interrupted.
1573 Note we do not turn off polling, because polling is only used
1574 when not interrupt_input, and thus not normally used on the systems
1575 which have this bug. On systems which use polling, there's no way
1576 to quit if polling is turned off. */
1578 /* Slow down polling. Some kernels have a bug which causes retrying
1579 connect to fail after a connect. */
1581 slow_down_interrupts ();
1585 /* A system call interrupted with a SIGALRM or SIGIO comes back
1586 here, with can_break_system_calls reset to 0. */
1587 SETJMP (break_system_call_jump);
1590 speed_up_interrupts ();
1592 /* In case something really weird happens ... */
1593 slow_down_interrupts ();
1596 /* Break out of connect with a signal (it isn't otherwise possible).
1597 Thus you don't get screwed with a hung network. */
1598 can_break_system_calls = 1;
1599 retval = connect (s, lres->ai_addr, lres->ai_addrlen);
1600 can_break_system_calls = 0;
1604 if (errno != EISCONN)
1608 if (errno == EADDRINUSE && retry < 20)
1610 /* A delay here is needed on some FreeBSD systems,
1611 and it is harmless, since this retrying takes time anyway
1612 and should be infrequent.
1613 `sleep-for' allowed for quitting this loop with interrupts
1614 slowed down so it can't be used here. Async timers should
1615 already be disabled at this point so we can use `sleep'. */
1625 speed_up_interrupts ();
1633 char servbuf[NI_MAXSERV];
1635 if (EQ (protocol, Qtcp))
1636 gni = getnameinfo (lres->ai_addr, lres->ai_addrlen,
1637 NULL, 0, servbuf, sizeof(servbuf),
1639 else /* EQ (protocol, Qudp) */
1640 gni = getnameinfo (lres->ai_addr, lres->ai_addrlen,
1641 NULL, 0, servbuf, sizeof(servbuf),
1642 NI_NUMERICSERV | NI_DGRAM);
1645 port = strtol (servbuf, NULL, 10);
1649 } /* address loop */
1651 speed_up_interrupts ();
1659 report_file_error ("connection failed", list2 (host, name));
1661 report_file_error ("error creating socket", list1 (name));
1663 #else /* ! HAVE_GETADDRINFO */
1664 struct sockaddr_in address;
1667 port = htons ((unsigned short) XINT (service));
1670 struct servent *svc_info;
1671 CHECK_STRING (service);
1673 if (EQ (protocol, Qtcp))
1674 svc_info = getservbyname ((char *) XSTRING_DATA (service), "tcp");
1675 else /* EQ (protocol, Qudp) */
1676 svc_info = getservbyname ((char *) XSTRING_DATA (service), "udp");
1679 error ("Unknown service \"%s\"", XSTRING_DATA (service));
1680 port = svc_info->s_port;
1683 get_internet_address (host, &address, ERROR_ME);
1684 address.sin_port = port;
1686 if (EQ (protocol, Qtcp))
1687 s = socket (address.sin_family, SOCK_STREAM, 0);
1688 else /* EQ (protocol, Qudp) */
1689 s = socket (address.sin_family, SOCK_DGRAM, 0);
1692 report_file_error ("error creating socket", list1 (name));
1694 /* Turn off interrupts here -- see comments below. There used to
1695 be code which called bind_polling_period() to slow the polling
1696 period down rather than turn it off, but that seems rather
1697 bogus to me. Best thing here is to use a non-blocking connect
1698 or something, to check for QUIT. */
1700 /* Comments that are not quite valid: */
1702 /* Kernel bugs (on Ultrix at least) cause lossage (not just EINTR)
1703 when connect is interrupted. So let's not let it get interrupted.
1704 Note we do not turn off polling, because polling is only used
1705 when not interrupt_input, and thus not normally used on the systems
1706 which have this bug. On systems which use polling, there's no way
1707 to quit if polling is turned off. */
1709 /* Slow down polling. Some kernels have a bug which causes retrying
1710 connect to fail after a connect. */
1712 slow_down_interrupts ();
1716 /* A system call interrupted with a SIGALRM or SIGIO comes back
1717 here, with can_break_system_calls reset to 0. */
1718 SETJMP (break_system_call_jump);
1721 speed_up_interrupts ();
1723 /* In case something really weird happens ... */
1724 slow_down_interrupts ();
1727 /* Break out of connect with a signal (it isn't otherwise possible).
1728 Thus you don't get screwed with a hung network. */
1729 can_break_system_calls = 1;
1730 retval = connect (s, (struct sockaddr *) &address, sizeof (address));
1731 can_break_system_calls = 0;
1732 if (retval == -1 && errno != EISCONN)
1737 if (errno == EADDRINUSE && retry < 20)
1739 /* A delay here is needed on some FreeBSD systems,
1740 and it is harmless, since this retrying takes time anyway
1741 and should be infrequent.
1742 `sleep-for' allowed for quitting this loop with interrupts
1743 slowed down so it can't be used here. Async timers should
1744 already be disabled at this point so we can use `sleep'. */
1752 speed_up_interrupts ();
1755 report_file_error ("connection failed", list2 (host, name));
1758 speed_up_interrupts ();
1759 #endif /* ! HAVE_GETADDRINFO */
1766 close (s); /* this used to be leaked; from Kyle Jones */
1767 report_file_error ("error duplicating socket", list1 (name));
1770 set_socket_nonblocking_maybe (inch, port, "tcp");
1772 *vinfd = (void*)inch;
1773 *voutfd = (void*)outch;
1777 #ifdef HAVE_MULTICAST
1779 /* Didier Verna <didier@xemacs.org> Nov. 28 1997.
1781 This function is similar to open-network-stream-internal, but provides a
1782 mean to open an UDP multicast connection instead of a TCP one. Like in the
1783 TCP case, the multicast connection will be seen as a sub-process,
1786 - Normally, we should use sendto and recvfrom with non connected
1787 sockets. The current code doesn't allow us to do this. In the future, it
1788 would be a good idea to extend the process data structure in order to deal
1789 properly with the different types network connections.
1790 - For the same reason, when leaving a multicast group, it is better to make
1791 a setsockopt - IP_DROP_MEMBERSHIP before closing the descriptors.
1792 Unfortunately, this can't be done here because delete_process doesn't know
1793 about the kind of connection we have. However, this is not such an
1798 unix_open_multicast_group (Lisp_Object name, Lisp_Object dest, Lisp_Object port,
1799 Lisp_Object ttl, void** vinfd, void** voutfd)
1802 struct sockaddr_in sa;
1803 struct protoent *udp;
1806 unsigned char thettl;
1807 int one = 1; /* For REUSEADDR */
1809 volatile int retry = 0;
1811 CHECK_STRING (dest);
1813 CHECK_NATNUM (port);
1814 theport = htons ((unsigned short) XINT (port));
1817 thettl = (unsigned char) XINT (ttl);
1819 if ((udp = getprotobyname ("udp")) == NULL)
1820 error ("No info available for UDP protocol");
1822 /* Init the sockets. Yes, I need 2 sockets. I couldn't duplicate one. */
1823 if ((rs = socket (PF_INET, SOCK_DGRAM, udp->p_proto)) < 0)
1824 report_file_error ("error creating socket", list1(name));
1825 if ((ws = socket (PF_INET, SOCK_DGRAM, udp->p_proto)) < 0)
1828 report_file_error ("error creating socket", list1(name));
1831 /* This will be used for both sockets */
1832 memset (&sa, 0, sizeof(sa));
1833 sa.sin_family = AF_INET;
1834 sa.sin_port = theport;
1835 sa.sin_addr.s_addr = htonl (inet_addr ((char *) XSTRING_DATA (dest)));
1837 /* Socket configuration for reading ------------------------ */
1839 /* Multiple connections from the same machine. This must be done before
1840 bind. If it fails, it shouldn't be fatal. The only consequence is that
1841 people won't be able to connect twice from the same machine. */
1842 if (setsockopt (rs, SOL_SOCKET, SO_REUSEADDR, (char *) &one, sizeof (one))
1844 warn_when_safe (Qmulticast, Qwarning, "Cannot reuse socket address");
1846 /* bind socket name */
1847 if (bind (rs, (struct sockaddr *)&sa, sizeof(sa)))
1851 report_file_error ("error binding socket", list2(name, port));
1854 /* join multicast group */
1855 imr.imr_multiaddr.s_addr = htonl (inet_addr ((char *) XSTRING_DATA (dest)));
1856 imr.imr_interface.s_addr = htonl (INADDR_ANY);
1857 if (setsockopt (rs, IPPROTO_IP, IP_ADD_MEMBERSHIP,
1858 (char *) &imr, sizeof (struct ip_mreq)) < 0)
1862 report_file_error ("error adding membership", list2(name, dest));
1865 /* Socket configuration for writing ----------------------- */
1867 /* Normally, there's no 'connect' in multicast, since we prefer to use
1868 'sendto' and 'recvfrom'. However, in order to handle this connection in
1869 the process-like way it is done for TCP, we must be able to use 'write'
1870 instead of 'sendto'. Consequently, we 'connect' this socket. */
1872 /* See open-network-stream-internal for comments on this part of the code */
1873 slow_down_interrupts ();
1877 /* A system call interrupted with a SIGALRM or SIGIO comes back
1878 here, with can_break_system_calls reset to 0. */
1879 SETJMP (break_system_call_jump);
1882 speed_up_interrupts ();
1884 /* In case something really weird happens ... */
1885 slow_down_interrupts ();
1888 /* Break out of connect with a signal (it isn't otherwise possible).
1889 Thus you don't get screwed with a hung network. */
1890 can_break_system_calls = 1;
1891 ret = connect (ws, (struct sockaddr *) &sa, sizeof (sa));
1892 can_break_system_calls = 0;
1893 if (ret == -1 && errno != EISCONN)
1899 if (errno == EADDRINUSE && retry < 20)
1901 /* A delay here is needed on some FreeBSD systems,
1902 and it is harmless, since this retrying takes time anyway
1903 and should be infrequent.
1904 `sleep-for' allowed for quitting this loop with interrupts
1905 slowed down so it can't be used here. Async timers should
1906 already be disabled at this point so we can use `sleep'. */
1914 speed_up_interrupts ();
1917 report_file_error ("error connecting socket", list2(name, port));
1920 speed_up_interrupts ();
1923 if (setsockopt (ws, IPPROTO_IP, IP_MULTICAST_TTL,
1924 (char *) &thettl, sizeof (thettl)) < 0)
1928 report_file_error ("error setting ttl", list2(name, ttl));
1931 set_socket_nonblocking_maybe (rs, theport, "udp");
1934 *voutfd = (void*)ws;
1937 #endif /* HAVE_MULTICAST */
1939 #endif /* HAVE_SOCKETS */
1942 /**********************************************************************/
1943 /* Initialization */
1944 /**********************************************************************/
1947 process_type_create_unix (void)
1949 PROCESS_HAS_METHOD (unix, alloc_process_data);
1950 PROCESS_HAS_METHOD (unix, mark_process_data);
1952 PROCESS_HAS_METHOD (unix, init_process);
1953 PROCESS_HAS_METHOD (unix, reap_exited_processes);
1955 PROCESS_HAS_METHOD (unix, init_process_io_handles);
1956 PROCESS_HAS_METHOD (unix, create_process);
1957 PROCESS_HAS_METHOD (unix, tooltalk_connection_p);
1958 PROCESS_HAS_METHOD (unix, set_window_size);
1960 PROCESS_HAS_METHOD (unix, update_status_if_terminated);
1962 PROCESS_HAS_METHOD (unix, send_process);
1963 PROCESS_HAS_METHOD (unix, process_send_eof);
1964 PROCESS_HAS_METHOD (unix, deactivate_process);
1965 PROCESS_HAS_METHOD (unix, kill_child_process);
1966 PROCESS_HAS_METHOD (unix, kill_process_by_pid);
1967 PROCESS_HAS_METHOD (unix, get_tty_name);
1969 PROCESS_HAS_METHOD (unix, canonicalize_host_name);
1970 PROCESS_HAS_METHOD (unix, open_network_stream);
1971 #ifdef HAVE_MULTICAST
1972 PROCESS_HAS_METHOD (unix, open_multicast_group);
1978 vars_of_process_unix (void)
1980 Fprovide (intern ("unix-processes"));
1983 #endif /* !defined (NO_SUBPROCESSES) */